nouveau SCHED_ERROR blocks keyboard

[joneall]

Occasionally (about once a day), i get an error either at logon just after booting or after screen lock (ctl-alt-L or screen saver). At such times, I cannot enter my password. Clicking with the mouse on the pw box shows no cursor and the keyboard is dead.

Here is syslog for just before and at the event.

Jul 1 11:06:48 jon-desktop colord[919]: (colord:919): Cd-WARNING **: failed to get session [pid 6560]: No such device or a
ddress
Jul 1 11:06:48 jon-desktop colord[919]: (colord:919): Cd-WARNING **: failed to get session [pid 6560]: No such device or a
ddress
Jul 1 11:07:04 jon-desktop org.kde.KScreen[1825]: kscreen: Primary output changed from KScreen::Output(Id: 175 , Name: "HD
MI-3" ) ( "HDMI-3" ) to KScreen::Output(Id: 175 , Name: "HDMI-3" ) ( "HDMI-3" )
Jul 1 11:08:32 jon-desktop kernel: [12607.489165] nouveau 0000:01:00.0: fifo: SCHED_ERROR 0a [CTXSW_TIMEOUT]
Jul 1 11:08:32 jon-desktop kernel: [12607.489171] nouveau 0000:01:00.0: fifo: sw engine fault on channel 12, recovering...
Jul 1 11:08:34 jon-desktop kernel: [12609.485944] nouveau 0000:01:00.0: fifo: runlist 0 update timeout
Jul 1 11:08:36 jon-desktop kernel: [12611.784039] nouveau 0000:01:00.0: fifo: SCHED_ERROR 0a [CTXSW_TIMEOUT]
Jul 1 11:08:40 jon-desktop kernel: [12616.078911] nouveau 0000:01:00.0: fifo: SCHED_ERROR 0a [CTXSW_TIMEOUT]

That last line repeats indefinitely until I boot again. A similar problem elsewhere says to add nomodeset or nouveau.modeset=0 to the grub params. I have never fiddled with them and would like to have some reason to hope that will help before I do.

Thanks in advance for any suggestions.

[GreyGeek]

Colord is the color pallet daemon and throws a lot of identical errors related to cups but doesn't hang my xserver.
If push came to shove I was going to do this:
https://wilkiecat.wordpress.com/2012...-ubuntu-12-04/

[joneall]

I had the pb again at my first boot this morning. The SCHED_ERR message does not occur in syslog. This should be a clear case, but...

It is not clear to me how to navigate in syslog. Where in it does a boot start? If I had that, I would know what occurred before. Why does it not insert a line right after boot saying something like "BOOT STARTING" so you can find it easily.

Since the date has changed, I can see the beginning of what corresponds to the bad boot. But I cannot find anything in there which repeats to show me where the second boot began, so I can see what happened just previously to that.

[GreyGeek]

Systemd-analyze is one good tool, journalctl is the second

Use
systemd-analyze
to get a short listing of kernel and user space boot times.
Use
systemd-analyze blame
to get a detailed listing. Replace blame with
plot >somename.svg
for a graph of blame's output.

journalctl
gives you not only what blame gives you but includes the plasma5 boot as well.
It will include a line saying "reboot" at a reboot point. More importantly, if you create the directory mentioned in this article:https://www.digitalocean.com/communi...e-systemd-logs
you can store boot logs and recall specific ones. That article explains it all!

[joneall]

I've had this pb twice today -- a new record. Nobody else seeing it?